Knowledge Integrated Machine Learning Techniques in Robotics and Game Playing

Machine-learning (ML) techniques are being used by materials researchers more and more, therefore it makes sense to look for lessons from other domains that are seeing comparable developments. We analyse the development of applied machine learning in robotics, gameplaying, and materials research in this review. We observe that the integration of ML in these domains happens in three steps: first, when it comes to specific hardware and software tools; second, when it comes to different stages in a workflow; and third, when it comes to the creation, generation, and representation of knowledge that can be applied to more than just one study known as knowledge integration. We find applicable insights from robots and gaming that can advance materials research. The application of gamification to concentrate community efforts on particular workflow integrations, set benchmarks, and create standard datasets, as well as the use of adaptive and accessible automation, are a few examples. Another is the use of hybrid algorithms, which combine data-based and model-based approaches to effectively handle complex tasks. We also highlight opportunities for interdisciplinary cooperation, like novel ways to represent and combine a wide range of knowledge as heuristics, physical laws, literature, and data with machine learning algorithms to produce fresh insights, and the responsible and fair application of technology for the good of society.